Children and Youth Services Review 32 (2010) 13171323

Contents lists available at ScienceDirect

Children and Youth Services Review

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c h i l d yo u t h

Effects of domestic violence on behavior problems of preschool-aged children: Do maternal mental health and parenting mediate the effects?
Chien-Chung Huang a,, Lih-Rong Wang b,1, Corinne Warrener a,2
a b

School of Social Work, Rutgers, the State University of New Jersey, 536 George Street, New Brunswick, NJ 08901, United States Department of Social Work, National Taiwan University, 1, Sec. 4, Roosevelt Road, Taipei, Taiwan

a r t i c l e

i n f o

a b s t r a c t
Using the rst four waves of the Fragile Families and Child Wellbeing Study, this article examined the effects of mothers who experienced domestic violence at Year 1 on the externalizing and internalizing behavior problems of children at Year 5 and investigated whether maternal mental health and parenting at Year 3 mediated those effects. Findings from structural equation modeling showed partial support for the hypothesized mediation effects. Consistent with the spillover hypothesis, domestic violence at Year 1 had a direct effect on maternal mental health at Year 3, which had direct effects on children's externalizing behavior problems at Year 5. Likewise, domestic violence at Year 1 had direct effects on parenting behavior and spanking at Year 3, and both parenting behavior and spanking then had direct effects on children's externalizing and internalizing behavior problems at Year 5. Domestic violence at Year 1 had direct effects on children's externalizing and internalizing behavior problems at Year 5. These results suggest that there are long-term effects of domestic violence on the behavior problems of preschool-aged children and that early interventions are needed to prevent later problems among children in families experiencing domestic violence. 2010 Elsevier Ltd. All rights reserved.

Article history: Received 5 January 2010 Received in revised form 26 April 2010 Accepted 26 April 2010 Available online 5 May 2010 Keywords: Domestic violence Children Externalizing behavior problems Internalizing behavior problems

1. Introduction The issue of domestic violence has been increasingly recognized as a persistent social and public health problem, particularly because of the tremendous, negative effects it has on victims' health and children's well-being (Carlson, McNutt, & Choi, 2003; Levendosky, Leahy, Bogat, Davidson, & von Eye, 2006; Lindhorst, Oxford, & Gillmore, 2007; Staggs, Long, Mason, Krishnan, & Riger, 2007; Zlotnick, Johnson, & Kohn, 2006). Although the effects of domestic violence on maternal mental health, parenting, and children's behavior problems have been well established in the literature, most studies have focused on school-age children (Edelson, 1999; Evans, Davies, & DiLillo, 2008; Levendosky & GrahamBermann, 2001; Sternberg et al., 1993), and only a few studies have examined the associations for children in early childhood (Carpenter & Stacks, 2009; Levendosky et al., 2006; Litrownik, Newton, Hunter, English, & Everson, 2003; Zerk, Mertin, & Proeve, 2009). Empirical evidence has clearly shown the importance of early childhood experiences to an individual's later development (Graces, Thomas, & Currie, 2002; Schweinhart, 2005; Walker, Greenwood, Hart, & Carta, 1994; West, Denton, & Germino-Hausken, 2000). Additionally, previous studies have found that there is a high prevalence of domestic

violence for low-income women (Catlett & Artis, 2004; Frias & Angel, 2007; Honeycutt, Marshall, & Weston, 2001; Kenney & McLanahan, 2006; Meisel, Chandler, & Rienzi, 2003; Tolman & Rosen, 2001) and that domestic violence tends to be frequent and increase over the early childhood years (Fantuzzo, Boruch, Beriama, Atkins, & Marcus, 1997; Huang, Son, & Wang, 2008; Layzer, Goodson, & Delange, 1985). This means that many preschool-aged children are at a high risk of experiencing domestic violence in their household. Given the large number of children being raised in low-income families and the many nancial challenges that these families face (Duncan & Brooks-Gunn, 1997; McLanahan et al., 2003), it is imperative to investigate the effects of domestic violence on the internalizing and externalizing behaviors for preschool-aged children because these outcomes have important effects on children's later development. To this end, this article uses the rst four waves of the Fragile Families and Child Wellbeing Study (FFCWS) to examine the effects of domestic violence on maternal mental health, parenting, and preschool-aged children's behavior problems. Increasing our understanding of the effects of domestic violence on these three outcomes will aid policymakers and frontline workers of domestic violence prevention programs in their efforts to design and implement intervention strategies for preschool-aged children who have experienced or are at risk of experiencing abuse. 2. Theoretic framework The bioecological theory of human development (Bronfenbrenner, 1979, 1986) underscores the importance of the dynamic interaction

Corresponding author. Tel.: +1 732 932 7520x117. E-mail address: huangc@rci.rutgers.edu (C.-C. Huang). 1 Tel.: +886 2 33661242. 2 Tel.: +1 732 932 7520x117. 0190-7409/$ see front matter 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.childyouth.2010.04.024

1318

C.-C. Huang et al. / Children and Youth Services Review 32 (2010) 13171323

between the child and the environment during a child's development. In bioecological theory, the child is the center of the developmental process, which is inuenced both through the child's own biology and a complex system of interconnected environmental layers (micro, meso, exo, and macrosystem). The microsystem is made up of networks in which the child has direct interpersonal relationships, such as the family and school. The mesosystem consists of the relationships between microsystem environments. The exosystem includes settings in which the child does not directly participate but that affect settings in which the child does routinely interact. The macrosystem is made up of larger reaching, less rigidly dened structures such as cultural values and social conditions that have the ability to inuence the child through the other environmental layers. In short, bioecological theory posits that the child is an active being who interacts bidirectionally with the societal context and that the child's development is affected by these interactions. Belsky (1984) took Bronfenbrenner's ecological framework a step further in the area of parenting and identied three domains of determinants of parenting. First, parenting may be inuenced by the characteristics of parents (e.g., parental psychological well-being), which are derived, in part, from their own developmental history. The second determinant is comprised of the characteristics of the child (e.g., gender and child temperament). The third domain is the social context within which the parent-child relationship is embedded. Contextual sources of stress (e.g., domestic violence) and support may promote or undermine parenting. In particular, Belsky proposed that, in buffering parent-child interactions from stress, a parent's psychological resources are more important than contextual sources of support, which in turn are more important than the characteristics of the child in predicting parenting behavior. Within the family microsystem, not only does domestic violence affect children's behavioral outcomes directly, but also it may have an indirect inuence through its effects on maternal psychological health and parenting. This paper draws on the spill-over hypothesis, essentially a direct application of the interrelatedness of bioecological theory. In particular, the spill-over hypothesis assumes that hostility and conict in one family system negatively inuence other family systems (Erel & Burman, 1995; Krisknakumar & Buehler, 2000; Levendosky et al., 2006). Similarly, the hypothesis also expects the negative effects of domestic violence on a mother's mental health and parenting to have a negative inuence on her children's behavior problems. 3. Empirical evidence Empirical studies, largely from school-aged children and infants, have supported an ecological model for explaining the effects of domestic violence on the well-being of mothers and their children. That is, domestic violence has negative effects on maternal mental health, reduces their ability to provide high-quality parenting, and results in poor behavioral consequences for children in the household (Anderson & Cramer-Benjamin, 1999; Coker et al., 2002; Edelson, 1999; Levendosky & Graham-Bermann, 2001; Levendosky et al., 2006). In terms of maternal mental health, victims have signicantly higher rates of both depression and anxiety (Coker et al., 2002; Levendosky & Graham-Bermann, 2001; Levendosky et al., 2006). Using population-based data from the National Violence Against Women Survey, Coker and colleagues found that about 29% of women (N = 6790) had experienced physical, sexual, or psychological domestic violence during their lifetime. Findings from regression analyses revealed that all three types of violence were associated with a high risk of depressive symptoms and chronic mental illness and that higher psychological violence scores were more strongly associated with these health outcomes than were physical violence scores (Coker et al., 2002). Levendosky and Graham-Bermann (2001) studied 120 women and one each of their children who were residing

in shelters or elsewhere in the community and found that domestic violence reduced maternal psychological functioning. Using 203 mother-infant dyads, Levendosky et al. (2006) found that past domestic violence, as well as current violence, signicantly affected maternal mental health and that the effects of both past and current violence were similar. Maternal mental health has been found to have direct and indirect effects on children's behavior via its effects on parenting. In terms of direct effects, studies have shown that poor maternal mental health is related to increased behavior problems for children (Jackson, BrooksGunn, Huang, & Glassman, 2000; Levendosky et al., 2006; Meadows, McLanahan, & Brooks-Gunn, 2007). For example, using the FFCWS (N = 2,120), Meadows et al. (2007) found that lower maternal mental health was associated with increased odds of anxious or depressed, attention decit, and oppositional deant disorders for children aged 3. Especially considering that paternal mental health had no signicant association with these problem behaviors, the ndings underscored the important inuence of maternal mental health on children's behavior problems. With respect to indirect effects, studies have shown that maternal mental health positively affects parenting quality (e.g., parental warmth) and that a high quality of parenting is associated with fewer behavior problems (Jackson et al., 2000; Levendosky et al., 2006). These ndings suggest that parenting may partially mediate the relationship between maternal mental health and children's behavior problems. With respect to parenting, domestic violence appears to reduce the quality of parenting (Anderson & Cramer-Benjamin, 1999; Levendosky & Graham-Bermann, 2001; Levendosky et al., 2006; Miller, Cowan, Cowan, Hetherington, & Clingempeel, 1993). For example, Miller and her colleagues (1993) used data from 41 families of the Becoming a Family Project and found that certain aspects of parenting quality, such as parental warmth, negatively affected child behavior problems, and the effects tended to be larger for children of preschool age than for adolescents. Likewise, both of the Levendosky studies (Levendosky & Graham-Bermann, 2001; Levendosky et al., 2006) indicated that domestic violence contributed to mothers being less able to respond warmly and sensitively to their children and less likely to bond positively with them. In their review of the literature, Anderson and Cramer-Benjamin (1999) found that interpersonal violence contributed to each parent being less emotionally and physically available to the child, and that reduced availability was likely to result in myriad internalizing and externalizing behavior problems. It is important to note that, in the studies reviewed, only positive dimensions of parenting (e.g., parental warmth towards the children) were measured. The lack of research measuring negative aspects of parenting, such as spanking, in studies analyzing the links between domestic violence and child behavior may be problematic given that these same studies have found that spanking is one of the most inuential factors in child behavior problems. In short, a number of empirical studies have indicated that domestic violence may negatively affect maternal mental health, which can then reduce parenting quality and lead to increased behavior problems for children. However, most of these studies have focused on school-aged children and infants and have used small or local datasets or both from the 1990s. Using data from a recent and large cohort, this paper aims to examine the effects of domestic violence on maternal mental health, parenting, and children's behavior problems during the preschool years, a time that has important effects on later development (Graces et al., 2002; Schweinhart, 2005; Walker et al., 1994; West, Denton, & GerminoHausken, 2000). For example, child development outcomes at age 5 are strong predictors of educational achievement in elementary school, which in turn is an important predictor of educational achievement, job earnings, and criminal activity in the adolescent and adult years (Graces et al., 2002; West, Denton, & GerminoHausken, 2000).

C.-C. Huang et al. / Children and Youth Services Review 32 (2010) 13171323

1319

4. Data and method 4.1. Data The data for this article come from The Fragile Families and Child Wellbeing Study (FFCWS), a longitudinal study designed to provide comprehensive information on the characteristics of unmarried parents and the well-being of their children. The data were collected in 20 U.S. cities with populations over 200,000 via stratied random sampling (see Reichman et al., 2001 for a detailed research design). Between 1998 and 2000, the initial core interviews were conducted with both the mother and father at the time of the baby's birth. The baseline data contained 4898 mothers (3711 unmarried and 1187 married). Follow-up core surveys were conducted when the child was 1, 3, and 5 years old. The rst four waves of surveys were used for this paper: baseline, Year 1, Year 3, and Year 5. Out of the 4898 eligible mothers at baseline, 4365 were interviewed at Year 1, 4231 at Year 3, and 4139 at Year 5. After the Year-3 core interview, families were asked to participate in an in-home assessment in which interviewers assessed the behaviors of the mothers and children and interviewed mothers about their parenting behaviors. For the Year-3 in-home assessment, 3254 mothers participated, and for the Year-5 in-home assessment, the number was 2977. A total of 2022 mothers completed all six data collections (the four core surveys and the two in-home assessments). Further, 1117 cases were dropped from the analysis due to incomplete information on the explanatory and dependent variables. The nal sample of this article included 905 mothers, who had complete information on all variables across all six surveys. The mean age of the nal sample was 25.4. About 49% of the sample was African American, and 25% and 23% were White and Hispanic, respectively. All other races made up the remaining 3% of the sample. For educational achievement, 29% of the sample did not complete a high-school education, 30% had a high school diploma, and 41% had more than a high-school education. There were no signicant differences in age, race, and education between mothers who did and did not experience domestic violence at Year 1. For example, the mean age of mothers who experienced domestic violence at Year 1 was 24.4, whereas the mean age was 25.5 for mothers who did not experience domestic violence at Year 1. To investigate the inuence of sample attrition on the ndings, baseline characteristics of the nal sample and the dropped cases were examined. The results showed that there were no differences by mother's age between the nal sample and the dropped cases; however, the dropped cases were signicantly more likely than the nal-sample cases to be Hispanic, to have less than a high-school education, and to report having a poor relationship quality with the father. Given these groups have been found to have higher household domestic violence rates and children's behavior problems than their counterparts in the literature, the sample attrition in the present study may have led to a downward bias in the estimates of the effects of domestic violence on child behavior problems reported in this study. 4.2. Measures 4.2.1. Child behavior problems Child behavior problems at Year 5 were measured in two dimensions: internalizing and externalizing problems. Internalizing behavioral problems are associated with poor mental health and social relationships, and externalizing behavioral problems are connected with healthdamaging behaviors such as bullying. The measures were taken from a partial scale of the Child Behavior Checklist for children aged 4-18 (CBCL/ 4-18; Achenbach, 1991). The CBCL/4-18 is a standardized set of measures for assessing the problematic behaviors of young children. Internalizing behavior problems included traits such as being withdrawn (9 items) or anxious or depressed (14 items), and externalizing problems included being aggressive (20 items) and delinquent behavior

(9 items). For the present study, the Cronbach's alphas for the internalizing and externalizing subscales were 0.75 and 0.85, respectively. For each subscale, the items were read to the mother, who was asked to indicate whether the statement was not true (0), sometimes or somewhat true (1), or very true or often true (2) of her child. Total scores for the internalizing and externalizing subscales were calculated by adding the scores for the items in that category. 4.2.2. Maternal mental health Maternal mental health at Year 3 was measured by the mother's experience of a depressive episode (1 = yes, 0 = no) based on the Composite International Diagnostic Interview, Short Form (CIDI-SF; Walters, Kessler, Nelson, & Mroczek, 2002). Mothers who reported feeling sad, blue, or depressed or who reported losing interest in most things for at least a 2-week period were subsequently asked seven questions related to those feelings, indicating whether they lost interest in most things, felt tired or lower in energy than usual, gained or lost 10 pounds without trying, had trouble falling asleep, had trouble concentrating, felt down, or had thoughts about death. Consistent with the CIDI-SF scoring method, mothers who reported experiencing three or more items were counted as having a major depressive episode. Alternatives to this strategy might be to use the number of indicators reported or the probability of major depression; we ran the analyses with both of these methods, but the results were not signicantly different from those reported in the text. 4.2.3. Parenting Parenting at Year 3 was measured in two dimensions: the interviewer's assessment of parenting behavior and the mother's report of using spanking as a disciplinary tactic. The assessment of parenting behavior included whether a mother was unresponsive to (6 items), harsh towards (5 items), or showed a lack of verbal or social skills with (3 items) the child during the home visit. The items were drawn from subscales of the Home Observation for Measurement of the Environment (HOME; Bradley, 1993). One point was assigned for each afrmative response, and the total number of afrmative responses was used to represent the level of parenting behavior. That is, the higher score, the better the parenting behavior. The total score ranged from 0 to 14, and the Cronbach's alpha of the scale was 0.79. Spanking was measured by the frequency with which the mother spanked the child when the child misbehaved or acted up in the month prior to the interview. Responses were coded into a ve-level variable: 0 (never), 1 (only once or), 2 (twice), 3 (a few times in the past month), 3 (a few times a week), and 4 (nearly every day). 4.2.4. Domestic violence Domestic violence was considered to be present at Year 1 if a woman reported that she experienced any amount of physical violence, emotional control, or sexual abuse at the hands of the child's father. The items he slapped or kicked you and he hit you with his st or a dangerous object were used to measure physical violence. Items measuring emotional control included he tried to isolate you from family and friends, he tried to prevent you from going to work and/or school, and he withheld money, made you ask for money, or took your money. The item he tried to make you have sex or do sexual things you didn't want to do was used to measure sexual abuse. There were three possible responses for each item: never, sometimes, and often. Because only a few mothers reported experiencing abuse often, the responses sometimes and often were combined into one category, labeled yes. Domestic violence was coded as yes if a woman had experienced any of above items and as no if a women reported never on all 6 items (1 = yes, 0 = no). 4.2.5. Child gender Given gender has been found to have strong effects on child behavior problems (Archer, 2004), gender at baseline was included as a control (1 = boy, 0 = girl).

1320

C.-C. Huang et al. / Children and Youth Services Review 32 (2010) 13171323

Table 1 Means, standard deviations, and correlations between variables. Variables 1. 2. 3. 4. 5. 6. 7.

+

Mean 0.09 0.15 12.63 0.95 12.59 5.35 0.51

SD 2.12 1.07 7.75 4.37

1 0.13 0.06+ 0.08 0.12 0.17 0.03

Domestic violence at year 1 (01) Depression at year 3 (01) Parenting behavior at year 3 (014) Spanking at year 3 (04) Externalizing behavior at year 5 (043) Internalizing behavior at year 5 (026) Male gender (01)

0.01 0.05 0.10 0.07 0.02

0.01 0.12 0.12 0.04

0.20 0.09 0.07

0.51 0.09

0.03

p b .10. p b .05. p b .01. p b .001.

4.3. Analytic techniques As described in the conceptual framework, the ecological model posits that children's behavior problems at Year 5 are partially determined by the domestic violence that the mother experienced at Year 1 and the presence (or absence) of maternal depression and the quality of parenting at Year 3. Specically, this paper investigates how domestic violence directly and indirectly operates to affect child behavior problems via its inuence on maternal mental health and parenting. That is, multiple equations are needed to examine the framework: two equations focusing on child behavior problems at Year 5 (internalizing and externalizing problems), two equations focusing on parenting at Year 3 (parenting behavior and spanking), and one equation on maternal mental health at Year 3. Thus, structural equation modeling, rather than OLS regressions, was used to test the conceptual framework. However, the number of variables in the equations had to be limited when estimating the equations jointly because otherwise the model would have been intractable. Thus, the structural equations included key variables that corresponded to the theoretical model. In results not shown, OLS regressions of child behavior problems were conducted with a rich set of maternal and child characteristics as controls, such as maternal age, race and ethnicity, educational attainment, substance abuse problems, relationship quality with the child's father, employment status, social support, and child temperament and health status. The OLS results of domestic violence on child behavior problems were not signicantly different from the results from the structural equation models. 5. Results Descriptive statistics and correlations of the variables are listed in Table 1. About 9% of mothers reported experiencing domestic violence

at Year 1 (n = 80). At Year 3, 15% of the sample reported experiencing a depressive episode. Overall, mothers reported positive parenting behaviors at Year 3. The mean of the behavior score was 12.6, with a range of 0 to 14. Likewise, the level of spanking at Year 3 was low, with a mean of 0.95 and a range of 0 to 4. The levels of externalizing and internalizing behavior problems were moderate at Year 5. The means for the externalizing and internalizing behavior problems were 12.6 (ranging from 0 to 43) and 5.4 (ranging from 0 to 26), respectively. About half of the children in the sample were boys. With respect to correlations between variables, as expected, domestic violence at Year 1 was signicantly and positively correlated with maternal depression at Year 3 (r = 0.13), spanking at Year 3 (r = 0.08), and externalizing (r = 0.12) and internalizing (r = 0.17) behavior problems at Year 5. Domestic violence at Year 1 was not signicantly correlated with parenting behavior at Year 3, although the direction of the correlation was negative as expected (r = 0.06). Mother's mental health at Year 3 was signicantly associated with children's externalizing and internalizing behavior problems at Year 5 (r = 0.10 and 0.07, respectively). However, depression was not correlated with parenting behavior or spanking. Positive parenting behavior at Year 3 was correlated with fewer externalizing and internalizing behavior problems at Year 5 (r = 0.12). Spanking at Year 3 was positively correlated with behavior problems at Year 5, and this relationship was stronger for externalizing (r = 0.20) than internalizing (r = 0.09) problems. The correlation between externalizing and internalizing behavior problems at Year 5 was 0.51. Male gender was correlated with spanking at Year 3 (r = 0.07) and externalizing behavior problems at Year 5 (r = 0.09). Fig. 1 shows the results of the structural equation models for the effects of domestic violence on the behavior problems of children. Assessments of model t were performed to determine whether the model depicted in Fig. 1 t the observed data. A Chi-square goodness-

Fig. 1. Structural equations modeling of domestic violence on behavior problems of children. N = 905, Chi-Square (7) = 9.9, p N .05; CFI = 0.99; RMSEA = 0.02. + p b .10, *p b .05, **p b .01, ***p b .001.

C.-C. Huang et al. / Children and Youth Services Review 32 (2010) 13171323

1321

of-t test was conducted. When testing the adequacy of a proposed model, a small, nonsignicant chi-square is evidence that the specied model is compatible with the observed data. The chi-square goodness-of-t test for the proposed model was nonsignicant (p N .05). Likewise, both the comparative t index (CFI) and the root mean square error of approximation (RMSEA) indicated that the model t the data well. The CFI value was 0.99, and the RMSEA was 0.03. With respect to coefcient estimates, the ndings indicate that domestic violence at Year 1 had direct effects on mothers' depression status (beta = 0.13, p b .001) and spanking (beta = 0.07, p b .05) at Year 3, as well as on children's externalizing (beta = 0.08, p b .01) and internalizing (beta = 0.15, p b .001) behavior problems at Year 5. Maternal depression at Year 3, however, did not have direct effects on parenting behavior or spanking at Year 3. Positive parenting behavior at Year 3 was directly related to fewer externalizing (beta = 0.12, p b .001) and internalizing (beta = 0.11, p b .001) behavior problems at Year 5, whereas spanking at Year 3 was associated with greater externalizing (beta = 0.19, p b .001) and internalizing (beta = 0.08, p b .001) behavior problems at Year 5. Male gender had direct effects on externalizing (beta = 0.07, p b .05) but not internalizing behavior problems at Year 5. Table 2 presents the indirect and total effects of domestic violence on children's behavior problems. The indirect effects of domestic violence at Year 1 on externalizing (beta = 0.03, p b .01) and internalizing (beta = 0.02, p b .01) behavior problems at Year 5 were signicant. In contrast, the indirect effects of maternal depression at Year 3 on children's behavior problems at Year 5 were not signicant. The results suggest that domestic violence at Year 1 affected spanking at Year 3 and that spanking inuenced children's externalizing and internalizing behavior problems at Year 5. In addition, the results suggest that domestic violence at Year 1 affected maternal depression at Year 3 and that depression had effects on children's externalizing behavior problems at Year 5. However, the link between maternal depression, parenting, and children's behavior problems was not supported by the data. 6. Discussion and conclusion This study examined the effects of domestic violence at Year 1 in the FFCWS data on the behavior problems of children at Year 5 and investigated whether maternal mental health and parenting at Year 3 mediated those effects. One of the strengths of this paper is that the data were observed longitudinally, across 5 years, which allowed for an examination of early domestic violence on later child outcomes.
Table 2 Decomposition of effects. Predictor Domestic violence Dependent variables Maternal depression Parenting behavior Spanking Externalizing behavior Internalizing behavior Parenting behavior Spanking Externalizing behavior Internalizing behavior Externalizing behavior Internalizing behavior Externalizing behavior Internalizing behavior Externalizing behavior Internalizing behavior

The ndings indicated that domestic violence at Year 1 had direct effects on both externalizing and internalizing behavior problems at Year 5. These results indicate that there are long-term effects for domestic violence and that early interventions are needed to prevent later problems among children in families experiencing domestic violence. The mediation effect occurs when the direct effect between the independent variable and the dependent variable reduces after controlling for the mediator variable. The mediation effect is complete when the direct effect is no longer signicant, and is partial when the direct effect is lower but still signicant, after controlling for the mediator variable. The ndings partially supported the hypotheses that mediation effects of maternal mental health and parenting at Year 3. First, consistent with the spillover hypothesis, domestic violence at Year 1 had a direct effect on maternal mental health at Year 3, which in turn had direct effects on children's externalizing behavior problems at Year 5. However, maternal mental health at Year 3 did not have direct effects on parenting at Year 3 or internalizing behavior problems at Year 5. This result provides evidence that maternal mental health partially mediated the effects of domestic violence on children's behavior problems in the studied data. Levendosky and her colleague (2006) also failed to nd an effect of maternal mental health on parenting, although previous studies have shown a link between these two variables (Levendosky & GrahamBermann, 2001; Lyons-Ruth, Wolfe, Lyubchik, & Steingard, 2002). The inconsistency of this nding over several studies points out the need for future studies to investigate the relationship between maternal mental health and parenting for this group of mothers and to examine the inuence of potential mediating variables such as parenting mastery. In the present study, domestic violence appeared to have effects on both maternal mental health and children's behavior problems. In addition, the fact that maternal mental health at Year 3 had direct effects on children's externalizing behavior problems at Year 5 suggests that maternal mental health may have effects on children's behavior via other factors rather than parenting. Further study is needed in this area. Second, domestic violence at Year 1 had direct effects on spanking at Year 3, and spanking at Year 3 then had direct effects on externalizing and internalizing behavior problems at Year 5. This is evidence that parenting behaviors partially mediated the effects of domestic violence on children's behavior problems. However, domestic violence at Year 1 had direct effects on externalizing and internalizing behavior problems at Year 5. Thus, spanking at Year 3 did not fully mediate the effects of domestic violence at Year 1 on children's behavior problems at Year 5. The direct effects of domestic violence on children's behavior problems merit discussion. Given that this study's model controlled for maternal mental health and parenting ability, the direct effects of domestic

Direct effect 0.13 0.06 + 0.07 0.08 0.15 0.00 0.04 0.07 0.05 0.12 0.11 0.19 0.08 0.07 0.02

Indirect effect 0.00 0.01 0.03 0.02 0.01 0.00

Total effect 0.13 0.06 0.08 0.12 0.17 +

Maternal depression

Parenting behavior Spanking Male gender

+

0.08 0.05 .12 0.11 0.19 0.08 0.07 0.02

p b .10. p b .05. p b .01. p b .001.

1322

C.-C. Huang et al. / Children and Youth Services Review 32 (2010) 13171323 Anderson, S. A., & Cramer-Benjamin, D. B. (1999). The impact of couple violence on parenting and children: an overview and clinical implications. The American Journal of Family Therapy, 27(1), 119. Archer, J. (2004). Sex differences in aggression in real-world settings: a meta-analytic review. Review of General Psychology, 8(4), 291322. Belsky, J. (1984). The determinants of parenting: a process model. Child Development, 55, 8396. Bradley, R. H. (1993). Children's home environments, health, behavior, and intervention efforts: a review using the HOME inventory as a marker measure. Genetic, Social, and General Psychology Monographs, 119, 437490. Bronfenbrenner, U. (1979). The Ecology of Human Development: Experiments by Nature and Design. Cambridge, MA: Harvard University Press. Bronfenbrenner, U. (1986). Ecology of the family as a context for human development: research perspectives. Developmental Psychology, 22, 723742. Carlson, B. E., McNutt, L., & Choi, D. Y. (2003). Childhood and adult abuse among women in primary health care. Journal of Interpersonal Violence, 18, 924941. Carpenter, G. L., & Stacks, A. M. (2009). Developmental effects of exposure to intimate partner violence in early childhood: a review of the literature. Children and Youth Services Review, 31, 831839. Catlett, B. S., & Artis, J. E. (2004). Critiquing the case for marriage promotion: how the promarriage movement misrepresents domestic violence research. Violence Against Women, 10(11), 12261244. Coker, A. L., Davis, K. E., Arias, I., Desai, S., Sanderson, M., Brandt, H. M., et al. (2002). Physical and mental health effects of intimate partner violence for men and women. American Journal of Preventive Medicine, 23(4), 260. Duncan, G. J., & Brooks-Gunn, J. (1997). Consequences of Growing Up Poor. New York: Russell Sage Foundation. Edelson, J. L. (1999). Children's witnessing of adult domestic violence. Journal of Interpersonal Violence, 14, 839. Erel, O., & Burman, B. (1995). Interrelatedness of marital relations and parent-child relations: a meta-analytic review. Psychological Bulletin, 118, 108132. Evans, S. E., Davies, C., & DiLillo, D. (2008). Exposure to domestic violence: a meta-analysis of child and adolescent outcomes. Aggression and Violent Behavior, 13(2), 131140. Fantuzzo, J. W., Boruch, R., Beriama, A., Atkins, M., & Marcus, S. (1997). Domestic violence and children: prevalence and risk in ve major U.S. cities. Journal of the American Academy of Child and Adolescent Psychiatry, 36, 116122. Frias, S. M., & Angel, R. J. (2007). Stability and change in the experience of partner violence among low-income women. Social Science Quarterly, 88(5), 12811306. Graces, E., Thomas, D., & Currie, J. (2002). Longer term effects of head start. American Economic Review, 92(4), 9991012. Graham-Bermann, S. A., Lynch, S., Banyard, V., DeVoe, E. R., & Halabu, H. (2007). Community-based intervention for children exposed to intimate partner violence: an efcacy trial. Journal of Consulting and Clinical Psychology, 75(2), 199209. Honeycutt, T. C., Marshall, L. L., & Weston, R. (2001). Toward ethnically specic models of employment, public assistance, and victimization. Violence Against Women, 7(2), 126140. Huang, C., Son, E., & Wang, L. (2008). Domestic Violence Among Unmarried Mothers: Prevalence, Determinants, and Implications for Marriage Promotion Policies. Paper presentation in the 2008 Annual Meeting of Association for Public Policy Analysis and Management, Los Angeles, CA, November 68. Jackson, A., Brooks-Gunn, J., Huang, C., & Glassman, M. (2000). Single mothers in lowwage jobs: nancial strain, parenting, and preschoolers' outcomes. Child Development, 71(5), 14091423. Kenney, C. T., & McLanahan, S. S. (2006). Why are cohabiting relationships more violent than marriage? Demography, 43(1), 127140. Krisknakumar, A., & Buehler, C. (2000). Interparental conict and parenting behaviors: a meta-analytic review. Family Relations, 49, 2544. Layzer, J. I., Goodson, B. D., & Delange, C. (1985). Children in shelters. Response to the Victimization of Women & Children, 9, 25. Levendosky, A. A., & Graham-Bermann, S. A. (2001). Parenting in battered women: the effects of domestic violence on women and their children. Journal of Family Violence, 16(2), 171. Levendosky, A. A., Leahy, K. L., Bogat, G. A., Davidson, W. S., & von Eye, A. (2006). Domestic violence, maternal parenting, maternal mental health and infant externalizing behavior. Journal of Family Psychology, 20(4), 544552. Lindhorst, T., Oxford, M., & Gillmore, M. R. (2007). Longitudinal effects of domestic violence on employment and welfare outcomes. Journal of Interpersonal Violence, 22 (7), 812828. Litrownik, A. J., Newton, R., Hunter, W. M., English, D., & Everson, M. D. (2003). Exposure to family violence in young at-risk children: a longitudinal look at the effects of victimization and witnessed physical and psychological aggression. Journal of Family Violence, 18, 5973. Lyons-Ruth, K., Wolfe, R., Lyubchik, A., & Steingard, R. (2002). Depressive symptoms in parents of children under age 3: Sociodemographic predictors, current correlates, and associated parenting behaviors. In N. Halfon, K. T. McLarn & M.A. Schuster (Eds.), Childrearing in America: Challenges facing parents with young children (pp. 217259). New York: Cambridge University Press. Meadows, S. O., McLanahan, S. S., & Brooks-Gunn, J. (2007). Parental depression and anxiety and early childhood behavior problems across family types. Journal of Marriage and Family, 69, 11621177. McLanahan, S. S., Garnkel, I., Reichman, N. E., Teitler, J. O., Carlson, M., & Audigier, C. N. (2003). The Fragile Families and Child Wellbeing Study: Baseline National Report. Princeton, NJ: Princeton University Center for Research on Child Wellbeing. Miller, N., Cowan, P. A., Cowan, C. P., Hetherington, E. M., & Clingempeel, W. G. (1993). Externalizing in preschoolers and early adolescents: a cross-study replication of a family model. Developmental Psychology, 29, 318.

violence on children's behavior problems imply that there were probably some other factors affecting those problems. A possible missing link is the father. Specically, the father could affect child outcomes through the overall father-child relationship, through the father's treatment or parenting of the child, or through the behavior he models for the child. It is also possible that the relationship dynamics between the mother and father have an impact. It is important to note that domestic violence at Year 1 did not have a signicant effect on the positive dimension of parenting (as measured by parenting behavior) or on the negative dimension of parenting (measured by spanking) at Year 3. Given that spanking is an important factor of child behavior problems in the literature, further study is warranted to investigate the effects of domestic violence on specic dimensions of parenting and to examine the potential variables mediating those relationships. The results show that domestic violence had negative impacts on both the mother and the child over an extended period of time. Although more research is necessary to fully understand the relationship between these variables, it is evident that early intervention is necessary in order to prevent later behavioral problems. Services and policies that affect families cannot ignore the fact that even children who may not be able to cognitively understand domestic violence can still be impacted by the existence of it within the family. Studies on domestic violence intervention programs for children have found that services provided to both the child and the mother at the same time were the most effective in mitigating the negative outcomes of domestic violence (Graham-Bermann, Lynch, Banyard, DeVoe, & Halabu, 2007; Sullivan, Egan, & Gooch, 2004). For example, Graham-Bermann and her colleagues (2007) studied 181 children aged 6 to 12 in a communitybased, domestic-violence intervention program. Three types of interventions were assessed: a child-only intervention, a child-plus-mother intervention, and a wait-list comparison. The results revealed that children in the child-plus-mother group showed the greatest improvement in behavior problems and attitudes about violence. Unfortunately, the intervention programs were only tested on school-aged children. Future research should examine the effectiveness of various interventions for preschool-aged children. In addition, regardless of which type of intervention would be most successful in lessening the negative impacts of domestic violence on children's behavior problems and other developmental outcomes, the present data make it clear that early intervention is important. The results should be interpreted within the context of two limitations. First, because of sample attrition across six surveys and incomplete data for the variables used in this paper, the nal sample only included about 20% of the original sample. This may limit the generalization of the ndings found in this paper. Second, the information on domestic violence, depression, and children's behavior problems was self-reported and thus is subject to reporting error, which, depending on the degree to which it was present, could have biased the estimates of the variables in the models. Despite these limitations, this paper increases our understanding of domestic violence on maternal mental health, parenting, and behavior problems for preschool-aged children. Acknowledgements Support for this article was provided by the Social Policy Research Center in National Taiwan University. Correspondence regarding this article can be sent to the rst author at huangc@rci.rutgers.edu or the second author at wanglr@ntu.edu.tw, or Social Policy Research Center, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei, 10617, Taiwan. References
Achenbach, T. M. (1991). Manual for the Child Behavior Checklist/4-18 and 1991 Prole. Burlington, VT: University of Vermont Department of Psychiatry.

C.-C. Huang et al. / Children and Youth Services Review 32 (2010) 13171323 Meisel, J., Chandler, D., & Rienzi, B. M. (2003). Domestic violence prevalence and effects on employment in two California TANF populations. Violence Against Women, 9 (10), 11911212. Reichman, N. E., Teitler, J. O., Garnkel, I., & McLanahan, S. S. (2001). Fragile families: Sample and design. Children and Youth Services Review, 23(4/5), 303326. Schweinhart, L. J. (2005). Lifetime Effects: The High/Scope Perry Preschool Study Through Age 40. The High/Scope Educational Research Foundation. Sullivan, M., Egan, M., & Gooch, M. (2004). Conjoint interventions for adult victims and children of domestic violence: a program evaluation. Research on Social Work Practice, 14, 163170. Staggs, S. L., Long, S. M., Mason, G. E., Krishnan, S., & Riger, S. (2007). Intimate partner violence, social support, and employment in the post-welfare reform era. Journal of Interpersonal Violence, 22(3), 345367. Sternberg, K. J., Lamb, M. E., Greenbaum, C., Cicchetti, D., Dawud, S., Cortes, R. M., et al. (1993). Effects of domestic violence on children's behavior problems and depression. Developmental Psychology, 29(1), 4452. Tolman, R. M., & Rosen, D. (2001). Domestic violence in the lives of women receiving welfare; mental health, substance dependence, and economic well-being. Violence Against Women, 7, 141158.

1323

Walker, D., Greenwood, C., Hart, B., & Carta, J. (1994). Prediction of school outcomes based on early language production and socioeconomic factors. Child Development, 65, 606662. Walters, E. E., Kessler, R. C., Nelson, C. B., & Mroczek, D. (2002). Scoring the World Health Organization's Composite International Diagnostic Interview Short Form (CIDI-SF). http://www.who.int/msa/cidi/CIDISFScoringMemo12-03-02.pdf. West, J., Denton, K., & Germino-Hausken, E. (2000). America's kindergarteners: Findings from the Early Childhood Longitudinal Study, Kindergarten Class of 199899, Fall 1998. National Center for Education Statistics (NCES) report 2000-070. Washington, DC: U.S. Department of Education, Ofce of Educational Research and Improvement. Zerk, D. M., Mertin, P. G., & Proeve, M. (2009). Domestic violence and maternal reports of young children's functioning. Journal of Family Violence, 24(7), 423432. Zlotnick, C., Johnson, D. M., & Kohn, R. (2006). Intimate partner violence and long-term psychosocial functioning in a national sample of American women. Journal of Interpersonal Violence, 21(2), 262275.